The epidemiology of recurrent Gram-negative bacteremia in a tertiary-care hospital Jonas Marschall a, ⁎ , Joshua Doherty b , David K. Warren a a Division of Infectious Diseases, Washington University School of Medicine, St. Louis, MO 63110, USA b Medical Informatics, BJC Healthcare, St. Louis, MO 63110, USA Received 19 August 2009; accepted 11 December 2009 Abstract We examined recurrent Gram-negative bacteremia in a prospective cohort from a tertiary hospital. Seventeen (7.1%) of 241 bacteremic patients developed recurrence (median time to recurrence = 44 days; range, 9–217 days). Recurrent and nonrecurrent bacteremic patients did not differ in clinical characteristics and mortality. © 2010 Elsevier Inc. All rights reserved. Keywords: Bacteremia; Gram-negative bacteria; Recurrence In the United States, approximately 250 000 episodes of bloodstream infections occur annually, causing substantial morbidity and mortality (Pittet and Wenzel, 1995). Recurrence of bloodstream infections is a recognized complication, particularly among Staphylococcus aureus bacteremias (Capdevila et al., 1994; Chang et al., 2003; Fowler et al., 1999; Johnson et al., 2003; Kreisel et al., 2006; Siegman-Igra et al., 2005), where it is associated with retained indwelling devices (Fowler et al., 1998; Johnson et al., 2003) and inferior antibiotic therapy (Chang et al., 2003; Fowler et al., 1999; Siegman-Igra et al., 2005). Gram-negative bacteremias account for approxi- mately 25% to 27% of bloodstream infections (Edmond et al., 1999; Wisplinghoff et al., 2004); however, little is known about the epidemiology of recurrent Gram-negative bacteremia (Mylotte and McDermott, 1988; Wendt et al., 1999b). Wendt et al. associated recurring Gram-negative bacteremia with intravascular catheters and an inadequate duration of antibiotic treatment. Mylotte and McDermott (1988) found an association between underlying malig- nancy and recurrence. Our objective was to prospectively determine the rate of recurrence of Gram-negative bloodstream infections in hospitalized patients and describe risk factors associated with recurrence. We performed a prospective cohort study of patients with Gram-negative bacteremia at Barnes-Jewish Hospital, St. Louis, MO, a 1250-bed teaching hospital, during a 6-month period between August 1, 2006, and January 31, 2007. We received daily electronic notification of patients with ≥1 blood culture positive for Gram-negative bacteria. Bacteremic adult patients housed in acute care inpatient wards were included. Bacteremic patients who were never admitted (e.g., positive blood culture from emergency department visit without admission) were excluded. Blood cultures were processed using BACTEC 9240 (Becton-Dickinson Diagnostic Systems, Sparks, MD). The disk-diffusion method was used for antibiotic susceptibility testing. Patient medical records were reviewed for demographics and medical history. Charlson comorbidity (Charlson et al., 1987) and McCabe severity of illness (McCabe and Jackson, 1962) scores were computed. Patients' vital signs and laboratory and pharmacy data were continuously reviewed during the admission. Recurrence of bacteremia was defined as repeated detection of Gram-negative bacteria in a blood culture after ≥1 negative blood culture and after an interval of N7 days, similar to Wendt et al. (1999b). Relapse (i.e., repeat bacteremia with the same strain) versus reinfection Available online at www.sciencedirect.com Diagnostic Microbiology and Infectious Disease 66 (2010) 456 – 459 www.elsevier.com/locate/diagmicrobio ⁎ Corresponding author. Tel.: +1-314-454-8225; fax: +1-314-454- 5392. E-mail address: jmarscha@dom.wustl.edu (J. Marschall). 0732-8893/$ – see front matter © 2010 Elsevier Inc. All rights reserved. doi:10.1016/j.diagmicrobio.2009.12.005